Difference Factor' KNN Collaborative Filtering Recommendation Algorithm

With the development of electronic commerce, Collaborative Filtering Recommendation system emerge, which uses machine learning algorithms for people provide a set of N items that will be of interest. In many user-based collaborative filtering applications based on KNN(K nearest neighbor algorithm), they only use similarity information(cosine similarity) between users, in some case, they have not use the difference information, so the precision and recall is not well. To address these problem, we propose a Difference Factor' K-NN collaborative filtering method, called DF-KNN. DF-KNN is an instance-based learning method and the key step in algorithms is how to use the difference factor and how to compute, the second step is mix similarity together. Our experimental evaluation on the MovieLens datasets show that the proposed DF-KNN and NDF-KNN(Normal Different Factor's KNN) are much efficient than the traditional user-neighborhood based KNN and provide recommendations whose quality is up to 13% better.